Sunday, January 22, 2006

Disturbing Infection Patterns in Turkey

Based on the Timeline published in Recombinomics: What's New I created the following chart of H5N1 infections in Turkey. Unfortunately it is a bit hard to read, but it can referenced against the text in the Recombinomics timeline (click on the image for an enlargement).

Red denotes known durations, or the onset dates of an illness when the duration is not known. In those cases when the end date of an individual's illness was not reported, I've used pink lines. A letter "D" at the end of a line indicates the individual died, while "R" indicates they recovered. If an individual's death was not reported I assume that they have recovered if a reasonable time has passed, but add a "?". I just use a "?" by itself if the timeframe suggests they are probably still in the hospital. Green squares note the dates of an event that is a likely point for the family to have contracted the infection.

The first key point to note is that the Ozcam family ate Christmas dinner with the Kocyigit family while the first Kocyigit son was ill but before the other children in that family developed symptoms(the upper greens square). Five days later they began to develop symptoms, fairly consistent with H5N1's normal incubation period.

The second point to notice is that the Kocyigit family caught the disease from birds, and had a severe mortality rate of 75%, very much in line with what we've come to expect from H5N1. Similarly the second Ozcam family caught the disease by slaughtering sick birds and had a 50% mortality rate, also consistent with other H5N1 outbreaks.

The first Ozcam family cluster is unique in two ways. One is that it is more than twice as large as any previous cluster we know of. The other is that not a single member of the Ozcam family died. Epidemiologists have long predicted that as H5N1 became better adapted to infecting humans it would become less lethal. The Ozcam family seems to be demonstrating this.

It isn't terribly obvious from the timeline above, but the linked clusters between the Kocyigit and Oscan families exhibit a disturbing pattern shown more clearly in the following graph.

Here the green line is the graph of an exponential growth function with a 5.5 day doubling period. Notice how closely the graph of the actual cases matches that function.

If the cases were caused by humans contracting the disease from an external source, then the trend should be a straight line as humans stumble into situations which lead them to be infected. It maight even tend to level off as humans learn to avoid the situations that lead them to become infected--which would be expected considering the massive education campagin and culling operations used in Turkey. Instead the cluster exhibits the classic exponential growth curve characteristic of unfettered population growth: 1 person infects 2 people who infect 4 more people, who infect 8 more people, and so on. It looks like a doubling rate of a little more than 5.5 days.

This suggests that the disease was not contracted by birds but was starting to be passed human to human in the ever expanding numbers that we would expect to see during a pandemic. What is hopeful is that quick hospitalization may have kept the virus from spreading outside the family. However even if disaster was averted this time around, the danger has not passed. If the strain of H5N1 currently circulating in birds has leaped the gap to sustained human to human transmission once, it will do it again, and again. We have to be lucky every time. H5N1 only needs to be lucky once.